A local training workbench for multiple models and engines
Prepare datasets, train models, and follow your runs in one place.
Download · Quick start · Documentation · 中文
Train Qwen-Image-2.1 LoRA with the DiffSynth engine, with training configuration and TOML preview side by side. Qwen-Image-2.1 beginner's guide (Chinese) · Explore the workspace.
- DiffSynth / Qwen-Image-2.1: BF16 text-to-image LoRA training with an isolated engine environment, image + TXT datasets, and previews during training. Image-edit training is not yet supported.
- Anima Fast: continue to train Anima 2.9B and T-LoRA with a dedicated runtime.
- Multiple engines, one workspace: manage Kohya, Anima Fast, Musubi, AI Toolkit, and DiffSynth from the same interface.
- Smoother training workflows: improved task management, configuration imports, and step/epoch handling.
Full changelog · Qwen-Image-2.1 guide (Chinese) · Anima Fast guide
Kohya · Anima Fast · Musubi · AI Toolkit · DiffSynth-Studio
Kohya is built in. Install the optional engines from Settings → Training engines, then select the appropriate engine on the training page. Available engines depend on your trainer version; not every model or feature supported upstream is integrated here.
For Qwen-Image-2.1, select DiffSynth-Studio → LoRA. Current support is BF16 text-to-image training, not image-edit training.
Qwen-Image-2.1 · Anima · SD 1.5 · SDXL · Flux · FLUX.2 Klein · Krea 2
Training targets, hardware requirements, and optional engine installation vary by model. See the training guides.
Windows users: download a portable package, extract it, and use the included launcher. An NVIDIA GPU is required; see setup and package selection.
Release status: 3.1.1 source is on main. Portable packages have their own release schedule; check the version and release notes on Releases. Older packages may not include DiffSynth or Qwen-Image-2.1 support.
For the 3.1.1 source version or Linux setup, follow run from source.